Date: Thu, 5 Aug 2004 15:35:27 GMT From: Mark Tinguely <tinguely@casselton.net> To: freebsd-gnats-submit@FreeBSD.org Subject: kern/70037: ALI AGP fails / panics system Message-ID: <200408051535.i75FZR0s093271@www.freebsd.org> Resent-Message-ID: <200408051540.i75FeVd6023450@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 70037
>Category: kern
>Synopsis: ALI AGP fails / panics system
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: freebsd-bugs
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Thu Aug 05 15:40:31 GMT 2004
>Closed-Date:
>Last-Modified:
>Originator: Mark Tinguely
>Release: FreeBSD 4.x and 5.x
>Organization:
>Environment:
>Description:
Many ALI based AGP chipsets cannot correct detect the Aperture size. In FreeBSD 4.x kernels, this causes a panic. In FreeBSD-current just fails to initialize the AGP. The problem is the ALI Aperture size test is wrong. This fix was taken from NetBSD. I do not agree with the mask in the agp_ali_detach() routine and put a "XXX comment". The below patch has been confirmed by another person who has the chipset.
>How-To-Repeat:
Boot a AGP enabled kernel on certain ALI chipset on FreeBSD 4.x or FreeBSD 5.x.
>Fix:
The following URL is a FreeBSD 4.10-RELEASE and FreeBSD-current (08/05/2004) unified diff to solve the problem:
http://www.casselton.net/~tinguely/agpali.patch.
This came from NetBSD. Visually, I think the agp_ali_detach() mask does not return the chipset to its original value.
>Release-Note:
>Audit-Trail:
>Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200408051535.i75FZR0s093271>